Although machine learning(ML)has emerged as a powerful tool for rapidly assessing grid contingencies,prior studies have largely considered a static grid topology in their analyses.This limits their application,since t...Although machine learning(ML)has emerged as a powerful tool for rapidly assessing grid contingencies,prior studies have largely considered a static grid topology in their analyses.This limits their application,since they need to be re-trained for every new topology.This paper explores the development of generalizable graph convolutional network(GCN)models by pre-training them across a wide range of grid topologies and contingency types.We found that a GCN model with auto-regressive moving average(ARMA)layers with a line graph representation of the grid offered the best predictive performance in predicting voltage magnitudes(VM)and voltage angles(VA).We introduced the concept of phantom nodes to consider disparate grid topologies with a varying number of nodes and lines.For pre-training the GCN ARMA model across a variety of topologies,distributed graphics processing unit(GPU)computing afforded us significant training scalability.The predictive performance of this model on grid topologies that were part of the training data is substantially better than the direct current(DC)approximation.Although direct application of the pre-trained model to topologies that are not part of the grid is not particularly satisfactory,fine-tuning with small amounts of data from a specific topology of interest significantly improves predictive performance.In the context of foundational models in ML,this paper highlights the feasibility of training large-scale GNN models to assess the reliability of power grids by considering a wide variety of grid topologies and contingency types.展开更多
With the emergence of new attack techniques,traffic classifiers usually fail to maintain the expected performance in real-world network environments.In order to have sufficient generalizability to deal with unknown ma...With the emergence of new attack techniques,traffic classifiers usually fail to maintain the expected performance in real-world network environments.In order to have sufficient generalizability to deal with unknown malicious samples,they require a large number of new samples for retraining.Considering the cost of data collection and labeling,data augmentation is an ideal solution.We propose an optimized noise-based traffic data augmentation system,ONTDAS.The system uses a gradient-based searching algorithm and an improved Bayesian optimizer to obtain optimized noise.The noise is injected into the original samples for data augmentation.Then,an improved bagging algorithm is used to integrate all the base traffic classifiers trained on noised datasets.The experiments verify ONTDAS on 6 types of base classifiers and 4 publicly available datasets respectively.The results show that ONTDAS can effectively enhance the traffic classifiers’performance and significantly improve their generalizability on unknown malicious samples.The system can also alleviate dataset imbalance.Moreover,the performance of ONTDAS is significantly superior to the existing data augmentation methods mentioned.展开更多
Jovani et al’s study contributes important evidence linking childhood trauma(CT)and parental socialization with at-risk mental state(ARMS)in non-clinical adolescents,demonstrating the mediating role of low levels of ...Jovani et al’s study contributes important evidence linking childhood trauma(CT)and parental socialization with at-risk mental state(ARMS)in non-clinical adolescents,demonstrating the mediating role of low levels of parental affection and communication in this relationship.This letter commends the study’s strengths while also identifying key issues that warrant further attention,including the limitations of cross-sectional design,potential perceptual biases,conceptual overlap between CT and parenting,and limited cultural generalizability.We advocate for longitudinal,culturally sensitive,and multi-informant approaches to further refine ARMS risk models,strengthen theoretical distinctions between CT and parenting,and inform targeted prevention strategies across diverse populations.We also extend the discussion by highlighting promising directions for future research.展开更多
Brain-computer interfaces(BCI)use neural activity as a control signal to enable direct communication between the human brain and external devices.The electrical signals generated by the brain are captured through elec...Brain-computer interfaces(BCI)use neural activity as a control signal to enable direct communication between the human brain and external devices.The electrical signals generated by the brain are captured through electroencephalogram(EEG)and translated into neural intentions reflecting the user’s behavior.Correct decoding of the neural intentions then facilitates the control of external devices.Reinforcement learning-based BCIs enhance decoders to complete tasks based only on feedback signals(rewards)from the environment,building a general framework for dynamic mapping from neural intentions to actions that adapt to changing environments.However,using traditional reinforcement learning methods can have challenges such as the curse of dimensionality and poor generalization.Therefore,in this paper,we use deep reinforcement learning to construct decoders for the correct decoding of EEG signals,demonstrate its feasibility through experiments,and demonstrate its stronger generalization on motion imaging(MI)EEG data signals with high dynamic characteristics.展开更多
Aim: The purpose of this study was to develop a scale, “parental anxiety about pediatric emergency medical care services” (PAPEMCS), and to evaluate its psychometric properties. Methods: Participants were 14,510 par...Aim: The purpose of this study was to develop a scale, “parental anxiety about pediatric emergency medical care services” (PAPEMCS), and to evaluate its psychometric properties. Methods: Participants were 14,510 parents with children 6 years old or younger in Kagawa Prefecture. Using each half of the participants, exploratory factor analysis was performed to generate items and factors for the PAPEMCS, and confirmatory factor analysis (CFA) was used to establish the construct validity. The generalizability of the PAPEMCS was evaluated by congruence tests and multigroup CFA. The usefulness of the PAPEMCS was established by the relationship between the PAPEMCS and non-urgent usage of pediatric emergency medical care services (PEMCS). Results: The PAPEMCS compromised 4 factors: “anxiety about quality of PEMCS”, “anxiety about PEMCS system”, “anxiety about public support”, and “anxiety about private support”. All reliability estimates (polychoric ordinal alpha coefficients, item-rest correlations), the item discrimination, 5 fit indices for CFA, the convergent validity (indicator reliabilities, composite reliabilities, average variance extracteds), and the discriminant validity fulfilled the acceptability thresholds. All generalizability estimates fulfilled the predetermined levels of acceptability (Tucker’s congruence coefficients, congruence tests, strict factorial invariance). The usefulness of the PAPEMCS was established by the higher scores of the PAPEMCS being related to non-urgent usage of PEMCS. Conclusions: The PAPEMCS demonstrated satisfactory reliability, validity, generalizability and usefulness. The PAPEMCS is useful to quantify the contents and extent of parental anxiety about PEMCS, and to clarify the mechanisms of non-urgent PEMCS usage.展开更多
The purpose of this study was to examine the burnout levels of research assistants in Ondokuz Mayis University and to examine the results of multiple linear regression model based on the results obtained from Maslach ...The purpose of this study was to examine the burnout levels of research assistants in Ondokuz Mayis University and to examine the results of multiple linear regression model based on the results obtained from Maslach Burnout Scale with Jackknife Method in terms of validity and generalizability. To do this, a questionnaire was given to 11 research assistants working at Ondokuz Mayis University and the burnout scores of this questionnaire were taken as the dependent variable of the multiple linear regression model. The variable of burnout was explained with the variables of age, weekly hours of classes taught, monthly average credit card debt, numbers of published articles and reports, gender, marital status, number of children and the departments of the research assistants. Dummy variables were assigned to the variables of gender, marital status, number of children and the departments of the research assistants and thus, they were made quantitative. The significance of the model as a result of multiple linear regressions was examined through backward elimination method. After this, for the five explanatory variables which influenced the variable of burnout, standardized model coefficients and coefficients of determination, and 95% confidence intervals of these values were estimated through Jackknife Method and the generalizability of the parameter estimation results of these variables on population was researched.展开更多
Artificial general intelligence (AGI) is the ability of an artificial intelligence (AI) agent to solve somewhat-arbitrary tasks in somewhat-arbitrary environments. Despite being a long-standing goal in the field of AI...Artificial general intelligence (AGI) is the ability of an artificial intelligence (AI) agent to solve somewhat-arbitrary tasks in somewhat-arbitrary environments. Despite being a long-standing goal in the field of AI, achieving AGI remains elusive. In this study, we empirically assessed the generalizability of AI agents by applying a deep reinforcement learning (DRL) approach to the medical domain. Our investigation involved examining how modifying the agent’s structure, task, and environment impacts its generality. Sample: An NIH chest X-ray dataset with 112,120 images and 15 medical conditions. We evaluated the agent’s performance on binary and multiclass classification tasks through a baseline model, a convolutional neural network model, a deep Q network model, and a proximal policy optimization model. Results: Our results suggest that DRL agents with the algorithmic flexibility to autonomously vary their macro/microstructures can generalize better across given tasks and environments.展开更多
This paper proposes a kernel-blending connection approximated by a neural network(KBNN)for image classification.A kernel mapping connection structure,guaranteed by the function approximation theorem,is devised to blen...This paper proposes a kernel-blending connection approximated by a neural network(KBNN)for image classification.A kernel mapping connection structure,guaranteed by the function approximation theorem,is devised to blend feature extraction and feature classification through neural network learning.First,a feature extractor learns features from the raw images.Next,an automatically constructed kernel mapping connection maps the feature vectors into a feature space.Finally,a linear classifier is used as an output layer of the neural network to provide classification results.Furthermore,a novel loss function involving a cross-entropy loss and a hinge loss is proposed to improve the generalizability of the neural network.Experimental results on three well-known image datasets illustrate that the proposed method has good classification accuracy and generalizability.展开更多
Advances in spinal cord injury-based research in the last 50 years have resulted in significant improvements to therapy options.However,the efficacy of such research could be further enhanced if threats to internal an...Advances in spinal cord injury-based research in the last 50 years have resulted in significant improvements to therapy options.However,the efficacy of such research could be further enhanced if threats to internal and external validity were addressed.To provide perspective,a sample topic was identified:the effects of acute and chronic exercise on clinical and sub-clinical markers of cardiovascular health.The intention was not a systematic review,nor a critique of exercise-based research,but rather a means to generate further discussion.Thirty-one articles were identified,and four common issues were found relating to:(1)sampling;(2)study design;(3)control group;and(4)clinical inference.These concerns were largely attributed to insufficient resources,and challenges associated with recruiting individuals with spinal cord injury.Overcoming these challenges will be difficult,but some opportunities include:(1)implementing multi-center trials;(2)sampling from subject groups appropriate to the research question;(3)including an appropriate control group;and(4)clearly defining clini-cal inference.These opportunities are not always feasible,and some easier to implement than others.However,addressing these concerns may assist in progressing spinal cord injury-based research,thereby helping to ensure steady advancement of therapy options for persons with spinal cord injury.展开更多
基金supported through the INL Laboratory Directed Research&Development(LDRD)Program under DOE Idaho Operations Office Contract DE-AC07-05ID14517This research made use of the resources of the High-Performance Computing Center at INL,which is supported by the U.S.Department of Energy’s Office of Nuclear Energy and the Nuclear Science User Facilities under contract no.DE-AC07-05ID14517.
文摘Although machine learning(ML)has emerged as a powerful tool for rapidly assessing grid contingencies,prior studies have largely considered a static grid topology in their analyses.This limits their application,since they need to be re-trained for every new topology.This paper explores the development of generalizable graph convolutional network(GCN)models by pre-training them across a wide range of grid topologies and contingency types.We found that a GCN model with auto-regressive moving average(ARMA)layers with a line graph representation of the grid offered the best predictive performance in predicting voltage magnitudes(VM)and voltage angles(VA).We introduced the concept of phantom nodes to consider disparate grid topologies with a varying number of nodes and lines.For pre-training the GCN ARMA model across a variety of topologies,distributed graphics processing unit(GPU)computing afforded us significant training scalability.The predictive performance of this model on grid topologies that were part of the training data is substantially better than the direct current(DC)approximation.Although direct application of the pre-trained model to topologies that are not part of the grid is not particularly satisfactory,fine-tuning with small amounts of data from a specific topology of interest significantly improves predictive performance.In the context of foundational models in ML,this paper highlights the feasibility of training large-scale GNN models to assess the reliability of power grids by considering a wide variety of grid topologies and contingency types.
基金supported in part by the National Key Research and Development Program of China(No.2022YFB4500800)the National Science Foundation of China(No.42071431).
文摘With the emergence of new attack techniques,traffic classifiers usually fail to maintain the expected performance in real-world network environments.In order to have sufficient generalizability to deal with unknown malicious samples,they require a large number of new samples for retraining.Considering the cost of data collection and labeling,data augmentation is an ideal solution.We propose an optimized noise-based traffic data augmentation system,ONTDAS.The system uses a gradient-based searching algorithm and an improved Bayesian optimizer to obtain optimized noise.The noise is injected into the original samples for data augmentation.Then,an improved bagging algorithm is used to integrate all the base traffic classifiers trained on noised datasets.The experiments verify ONTDAS on 6 types of base classifiers and 4 publicly available datasets respectively.The results show that ONTDAS can effectively enhance the traffic classifiers’performance and significantly improve their generalizability on unknown malicious samples.The system can also alleviate dataset imbalance.Moreover,the performance of ONTDAS is significantly superior to the existing data augmentation methods mentioned.
文摘Jovani et al’s study contributes important evidence linking childhood trauma(CT)and parental socialization with at-risk mental state(ARMS)in non-clinical adolescents,demonstrating the mediating role of low levels of parental affection and communication in this relationship.This letter commends the study’s strengths while also identifying key issues that warrant further attention,including the limitations of cross-sectional design,potential perceptual biases,conceptual overlap between CT and parenting,and limited cultural generalizability.We advocate for longitudinal,culturally sensitive,and multi-informant approaches to further refine ARMS risk models,strengthen theoretical distinctions between CT and parenting,and inform targeted prevention strategies across diverse populations.We also extend the discussion by highlighting promising directions for future research.
文摘Brain-computer interfaces(BCI)use neural activity as a control signal to enable direct communication between the human brain and external devices.The electrical signals generated by the brain are captured through electroencephalogram(EEG)and translated into neural intentions reflecting the user’s behavior.Correct decoding of the neural intentions then facilitates the control of external devices.Reinforcement learning-based BCIs enhance decoders to complete tasks based only on feedback signals(rewards)from the environment,building a general framework for dynamic mapping from neural intentions to actions that adapt to changing environments.However,using traditional reinforcement learning methods can have challenges such as the curse of dimensionality and poor generalization.Therefore,in this paper,we use deep reinforcement learning to construct decoders for the correct decoding of EEG signals,demonstrate its feasibility through experiments,and demonstrate its stronger generalization on motion imaging(MI)EEG data signals with high dynamic characteristics.
文摘Aim: The purpose of this study was to develop a scale, “parental anxiety about pediatric emergency medical care services” (PAPEMCS), and to evaluate its psychometric properties. Methods: Participants were 14,510 parents with children 6 years old or younger in Kagawa Prefecture. Using each half of the participants, exploratory factor analysis was performed to generate items and factors for the PAPEMCS, and confirmatory factor analysis (CFA) was used to establish the construct validity. The generalizability of the PAPEMCS was evaluated by congruence tests and multigroup CFA. The usefulness of the PAPEMCS was established by the relationship between the PAPEMCS and non-urgent usage of pediatric emergency medical care services (PEMCS). Results: The PAPEMCS compromised 4 factors: “anxiety about quality of PEMCS”, “anxiety about PEMCS system”, “anxiety about public support”, and “anxiety about private support”. All reliability estimates (polychoric ordinal alpha coefficients, item-rest correlations), the item discrimination, 5 fit indices for CFA, the convergent validity (indicator reliabilities, composite reliabilities, average variance extracteds), and the discriminant validity fulfilled the acceptability thresholds. All generalizability estimates fulfilled the predetermined levels of acceptability (Tucker’s congruence coefficients, congruence tests, strict factorial invariance). The usefulness of the PAPEMCS was established by the higher scores of the PAPEMCS being related to non-urgent usage of PEMCS. Conclusions: The PAPEMCS demonstrated satisfactory reliability, validity, generalizability and usefulness. The PAPEMCS is useful to quantify the contents and extent of parental anxiety about PEMCS, and to clarify the mechanisms of non-urgent PEMCS usage.
文摘The purpose of this study was to examine the burnout levels of research assistants in Ondokuz Mayis University and to examine the results of multiple linear regression model based on the results obtained from Maslach Burnout Scale with Jackknife Method in terms of validity and generalizability. To do this, a questionnaire was given to 11 research assistants working at Ondokuz Mayis University and the burnout scores of this questionnaire were taken as the dependent variable of the multiple linear regression model. The variable of burnout was explained with the variables of age, weekly hours of classes taught, monthly average credit card debt, numbers of published articles and reports, gender, marital status, number of children and the departments of the research assistants. Dummy variables were assigned to the variables of gender, marital status, number of children and the departments of the research assistants and thus, they were made quantitative. The significance of the model as a result of multiple linear regressions was examined through backward elimination method. After this, for the five explanatory variables which influenced the variable of burnout, standardized model coefficients and coefficients of determination, and 95% confidence intervals of these values were estimated through Jackknife Method and the generalizability of the parameter estimation results of these variables on population was researched.
文摘Artificial general intelligence (AGI) is the ability of an artificial intelligence (AI) agent to solve somewhat-arbitrary tasks in somewhat-arbitrary environments. Despite being a long-standing goal in the field of AI, achieving AGI remains elusive. In this study, we empirically assessed the generalizability of AI agents by applying a deep reinforcement learning (DRL) approach to the medical domain. Our investigation involved examining how modifying the agent’s structure, task, and environment impacts its generality. Sample: An NIH chest X-ray dataset with 112,120 images and 15 medical conditions. We evaluated the agent’s performance on binary and multiclass classification tasks through a baseline model, a convolutional neural network model, a deep Q network model, and a proximal policy optimization model. Results: Our results suggest that DRL agents with the algorithmic flexibility to autonomously vary their macro/microstructures can generalize better across given tasks and environments.
基金the National Natural Science Foundation of China(Grant Nos.61972227 and 61672018)the Natural Science Foundation of Shandong Province(Grant No.ZR2019MF051)+1 种基金the Primary Research and Development Plan of Shandong Province(Grant No.2018GGX101013)the Fostering Project of Dominant Discipline and Talent Team of Shandong Province Higher Education Institutions。
文摘This paper proposes a kernel-blending connection approximated by a neural network(KBNN)for image classification.A kernel mapping connection structure,guaranteed by the function approximation theorem,is devised to blend feature extraction and feature classification through neural network learning.First,a feature extractor learns features from the raw images.Next,an automatically constructed kernel mapping connection maps the feature vectors into a feature space.Finally,a linear classifier is used as an output layer of the neural network to provide classification results.Furthermore,a novel loss function involving a cross-entropy loss and a hinge loss is proposed to improve the generalizability of the neural network.Experimental results on three well-known image datasets illustrate that the proposed method has good classification accuracy and generalizability.
文摘Advances in spinal cord injury-based research in the last 50 years have resulted in significant improvements to therapy options.However,the efficacy of such research could be further enhanced if threats to internal and external validity were addressed.To provide perspective,a sample topic was identified:the effects of acute and chronic exercise on clinical and sub-clinical markers of cardiovascular health.The intention was not a systematic review,nor a critique of exercise-based research,but rather a means to generate further discussion.Thirty-one articles were identified,and four common issues were found relating to:(1)sampling;(2)study design;(3)control group;and(4)clinical inference.These concerns were largely attributed to insufficient resources,and challenges associated with recruiting individuals with spinal cord injury.Overcoming these challenges will be difficult,but some opportunities include:(1)implementing multi-center trials;(2)sampling from subject groups appropriate to the research question;(3)including an appropriate control group;and(4)clearly defining clini-cal inference.These opportunities are not always feasible,and some easier to implement than others.However,addressing these concerns may assist in progressing spinal cord injury-based research,thereby helping to ensure steady advancement of therapy options for persons with spinal cord injury.
